Russ Feingold is really something else. He's a Rhode Scholar, so he's obviously pretty smart. However, he is arguably the least powerful senator and probably won't last another term. Along with McCain, he charged ahead with the Campaign Finance Reform bill. That is why he is the least powerful man because every large corporation doesn't want their voice in congress to disappear. Than and most senators realize that whats keeping them in office is the large sums of money they receive.
The most important aspect missed in this comparison is the value of Windows experience to the students who will ultimately be using these systems. Knowing how to use Microsoft Word proficiently will be a much better asset and skill than being proficient in OpenOffice. This unfortunately isn't going to change in the near future. Consider a college graduate applying for an accountant job. I guarantee you that not having much experience in Microsoft Excel would put an applicant at a disadvantage.
RISC vs. CISC doesn't matter anymore. Both AMD and Intel convert the CISC architectural instructions into RISC-like internal instructions called "micro-ops".
The Iron Law of microprocessor performance states:
time = Instructions/Program * Cycles/Instruction * Seconds/Cycle
Notice that the above cancels to Seconds/Program.
So given the same program, you can do one of two things to reduce the execution time:
1. Reduce Cycles/Instruction. This is done by executing more instructions at a time (Superscalar execution), and reducing the penalties of speculation (predicting what the program will do).
2. Reducing the Seconds/Cycle (increasing clock rate). This is the approach Intel has taken with Netburst (Pentium 4).
By making the PIV pipe 22 stages long, this helps Intel increase the clock rate. It also helps throughput. Assuming an ideal world where no data dependencies exist, more stages == better throughput. However, Control and Data Hazards always messes up a pipeline and the longer the pipeline, the more of a penalty on a branch misprediction. This increases cycles/instruction (CPI).
The average CPI of AMD and PIII is much lower than the P4, however it makes up for this in clock rate.
And in other news, Bram Moolenaar announced that the upcoming version of Vim will be released as version 23. During a recent interview, Bram stated that "those Emacs morons think they can gain market share by inflating the version number. This jump in Vim versioning merely helps consumers accurately choose the best text editor. With Vim v6.0, some uninformed consumers may believe that Vim does not have as many features as Emacs v21. Besides, kudos to Michael Jordan for making another comeback...just like vi!".
Richard Stallman could not be reached for comment. Sources believe that he is in Afghanistan promoting the name "GNU/Emacs" instead of just "Emacs".
I've tried all the major versions of KDE. Sure is nice...my grandma could probably use it. Too cluttered and slow for me. I've also tried all the major versions of Gnome. Seems a little zippier than KDE, and my mom could probably use it. Too much clutter for me.
I keep going back to Blackbox everytime I try a new "Desktop Environment".
Yes, the new IBM Regatta processors just announced last week have the fasted SPECint and SPECfp perfomance. They dethrone the 2GHz Pentium 4 as the SPECint king, and the lastest Alpha (21???) as the SPECfp king.
Most people I know, who've had cellphone service for several years, use a 2-year old mobile.
CDMA cell sizes can be huge (i'm talking CDMA in generic terms here). The capacity is just decreased because the noise floor is higher. Remember this is a spread spectrum technology. The U.S. military uses CDMA technology for talking to fighter jets. Their data rate is so low (1 kb/s instead of 14.4 kb/s) that the noise flow can be extremely high thus they have "cell sites" that are hundreds of miles wide. In CDMA, you can have a tradeoff between cell size, users, and data rate. Since the data rate is fixed (sort-of), that leaves a tradeoff between capacity and cell size (noise floor)
Sorry, I don't have my CDMA theory book in front of me, and I don't know the equations off the top of my head. If I get the time, I will look them up for you.
Can anyone else explain to this guy why you can do this in CDMA? Its been too long since I've studied the stuff to get the detail he wants.
I disagree with point 2. IS-95 interoperates better with AMPS. You can have an IS-95 network running on the A frequencies, and standard AMPS on the B frequencies. I'm not sure of this, but it may even be possible to run AMPS and IS-95 in A or B frequencies. The 1.25MHz spectrum of IS-95 is evenly divisable by the channel frequencies of individual AMPS channels.
Many carriers in the U.S. will evolve to CDMA2000, which can be regarded as a stepping stone to WCDMA as soon as they figure out the spectrum. Kind of like how GPRS is a stepping stone from GSM to UMTS.
How long as that nifty Ericsson phone been around? I'm sure its a great phone, but realize that carriers have to plan their networks years before all of these nifty phones come to market.
I'm not knocking GSM...its a great standard and Europe is very fortunate to have a nice uniform standard. It just wasn't right for the U.S. at the time (early 90's). I also like the SIM cards which are non-existant in IS-95.
Does you phone work in central Wyoming, Wisconsin, Montana, Idaho, Arkansas, Oklahoma, New Mexico, or any of the other sparcely populated areas? I guess if you never go to these places, then who cares, huh?
GSM is, and still isn't right for the United States for 2 reasons.
1) The European GSM standard uses 900MHz and 1800MHz. Those frequencies are used by the United States government, and have been long before GSM came around. So that's why GSM in the U.S. uses different frequencies. So any GSM phone that is, say 2 years old, won't work in Europe and vice versa.
2) Optimal cell size is a function of population density. Digital technologies, especially GSM, require smaller cell sizes. Simply put, places like Wyoming are not going to be getting digital anytime soon. But they do have analog because you can make those cells huge. GSM does not interoperate well with AMPS. CDMA IS-95 does. You can run a CDMA network and an AMPS network at 800MHz.
I also believe that CDMA is the future (not IS-95). Sure, UMTS is based on time-division, but the Docomo 3G call stack is CDMA-based.
So go buy your VoiceStream phone if you live in a big city. I live in Chicago. VoiceStream is there, but no way will I get a VoiceStream phone because if I ever go to Wisconsin, my phone will not get service. My Verizon CDMA phone gets service anywhere in the country.
At least where I received my B.S. in Electrical Engineering. My senior year, it was like high school all over. All of my classes where in the same building!! In fact, sometimes I didn't even change rooms. No, my school wasn't small being its Big-Ten with over 40,000 students.
So then I went to work for 2 years for a Fortune 50 company and realized that a B.S. in Engineering is just grunt work for several years until you become a manager. So now I'm getting my M.S. or PhD in Computer Science.
Do I wish I had a more well-rounded educations? Hell yes. And I plan to make the most out of Graduate School by taking some off-the-wall classes like Art, Economics, Political Science, and Sociology.
It has been shown that the sweet spot for R&D spending is between 5-15%. Companies that spend more money on R&D inevitable fail. I pulled this statement straight out of Patterson & Hennessey's unpublished third addition of "A Quantitative Approach to Computer Architecture"
Intel is reknown for their manufacturing prowess. Of course the P4 has a low yield. They are just ramping up the Net-burst architecture. Athlon has been running for awhile now.
Personally, I would buy AMD if they could make a chip that doesn't burn up if the CPU fan goes. An Athlon 1GHz uses more than twice the wattage as a 1GHz PIII. Amortise the electricity savings of using a PIII 24/7 for 5 years into the price difference...it may surprise you.
AMD really doesn't stand a chance unless Itanium (IA-64) is a disaster. Intel is betting everything on 64-bit. AMD will never be able to make a chip that uses the Itanium ISA (instruction set architecture). AMD is working on 64-bit extensions to IA-32. However, it will be pointless if IA-64 becomes the IA-32 of the 80's and 90's.
The ability to overclock a chip says nothing about the quality. It might mean that AMD doesn't have their statistical quality under as tight as control as Intel. They may design a chip to run at 1.6GHz, but because of process inconsistencies, they can only sell them as 1GHz chips. So I ask, are all Athlon's the same? Can every Athlon be clocked the same way or is it a random distribution?
That's basically how it works. A PIII-650 and a PIII-1ghz are the same design, same process, and same chip. However, the more processors they manufacture, the better they get at finetuning their manufacturing/process.
I suggest reading Hennessy & Patterson's "Quantitative Approach to Computer Architecture".
The ability to overclock a chip says nothing about the quality. It might mean that AMD doesn't have their statistical quality under as tight as control as Intel. They may make 1.6GHz chips, but because of process inconsistencies, they can only sell them as 1GHz chips. So I ask, are all Athlon's the same? Can every Athlon be clocked the same way or is it a random distribution?
The energy required per transistor is proportional to the product of the load capacitance of the transistor, the frequency of switching, and the square of the voltage.
Can you clarify how decreasing the frequency allows you to lower the voltage? I don't have my VLSI book in front of me right now, but I thought that voltage was proportional to the gate distance...thus smaller processes can operate at a lower voltage.
Well, I don't know what your definition of a mile is. A 1GHz Athlon is faster than a 1GHz PIII, but not by an order of magnitude. I don't have numbers in front of me. Is the SPECint maybe 15% better?
Slowing the clock would not dramatically affect power consumption. Sure, it would help. The Athlon literally uses twice as much power thus dissipating twice as much heat.
So, little boy, when your Athlon burns up because some dust clogged your cooling fan, I'll be happily downloading p0rn with my Intel;)
You're right! G3/G4's have a superior architecture. I just wish they were commodity hardware like x86...
The x86 has its roots from the 8008 designed in 1974. In fact, the designers of the 8086 spent only a few weeks kludging together an instruction set. Software compatibility means that x86 is still is shitty architecture.
And since the average Joe Blow thinks that MHz is everything, Intel is pushing the P4 to 2GHz even though the SPECint/MHz is far less than the PIII and G4.
Slashdot Mirror
Russ Feingold is really something else. He's a Rhode Scholar, so he's obviously pretty smart. However, he is arguably the least powerful senator and probably won't last another term. Along with McCain, he charged ahead with the Campaign Finance Reform bill. That is why he is the least powerful man because every large corporation doesn't want their voice in congress to disappear. Than and most senators realize that whats keeping them in office is the large sums of money they receive.
The most important aspect missed in this comparison is the value of Windows experience to the students who will ultimately be using these systems. Knowing how to use Microsoft Word proficiently will be a much better asset and skill than being proficient in OpenOffice. This unfortunately isn't going to change in the near future. Consider a college graduate applying for an accountant job. I guarantee you that not having much experience in Microsoft Excel would put an applicant at a disadvantage.
RISC vs. CISC doesn't matter anymore. Both AMD and Intel convert the CISC architectural instructions into RISC-like internal instructions called "micro-ops".
The Iron Law of microprocessor performance states:
time = Instructions/Program * Cycles/Instruction * Seconds/Cycle
Notice that the above cancels to Seconds/Program.
So given the same program, you can do one of two things to reduce the execution time:
1. Reduce Cycles/Instruction. This is done by executing more instructions at a time (Superscalar execution), and reducing the penalties of speculation (predicting what the program will do).
2. Reducing the Seconds/Cycle (increasing clock rate). This is the approach Intel has taken with Netburst (Pentium 4).
By making the PIV pipe 22 stages long, this helps Intel increase the clock rate. It also helps throughput. Assuming an ideal world where no data dependencies exist, more stages == better throughput. However, Control and Data Hazards always messes up a pipeline and the longer the pipeline, the more of a penalty on a branch misprediction. This increases cycles/instruction (CPI).
The average CPI of AMD and PIII is much lower than the P4, however it makes up for this in clock rate.
And in other news, Bram Moolenaar announced that the upcoming version of Vim will be released as version 23. During a recent interview, Bram stated that "those Emacs morons think they can gain market share by inflating the version number. This jump in Vim versioning merely helps consumers accurately choose the best text editor. With Vim v6.0, some uninformed consumers may believe that Vim does not have as many features as Emacs v21. Besides, kudos to Michael Jordan for making another comeback...just like vi!".
Richard Stallman could not be reached for comment. Sources believe that he is in Afghanistan promoting the name "GNU/Emacs" instead of just "Emacs".
I've tried all the major versions of KDE. Sure is nice...my grandma could probably use it. Too cluttered and slow for me. I've also tried all the major versions of Gnome. Seems a little zippier than KDE, and my mom could probably use it. Too much clutter for me.
I keep going back to Blackbox everytime I try a new "Desktop Environment".
Gee, the ALU of a 2GHz Intel PIV is double-pumped meaning it runs at 4Ghz!! They should call it the Pentium 4000!!
And where do Linus and Miguel reside now?
That's right, the U.S.A.
This is a stupid post?
I live in a city of 150,000. Just signed up for DSL yesterday. I had 3 local choices for DSL (not Cable though).
Broadband is not dead where I live (Wisconsin). Shit, my 65-year old father has DSL and he lives in a town of 8,000!!!
allows a bug like this? I was about to upgrade my Kernel to 2.4.11...good thing I hestitated.
This release process, along with the 2.4.x VM problems may induce me to try FreeBSD....
Yes, the new IBM Regatta processors just announced last week have the fasted SPECint and SPECfp perfomance. They dethrone the 2GHz Pentium 4 as the SPECint king, and the lastest Alpha (21???) as the SPECfp king.
This is like saying "Joe Blow's toilet flushes faster than Joe Fish's toilet"
Most people I know, who've had cellphone service for several years, use a 2-year old mobile.
CDMA cell sizes can be huge (i'm talking CDMA in generic terms here). The capacity is just decreased because the noise floor is higher. Remember this is a spread spectrum technology. The U.S. military uses CDMA technology for talking to fighter jets. Their data rate is so low (1 kb/s instead of 14.4 kb/s) that the noise flow can be extremely high thus they have "cell sites" that are hundreds of miles wide. In CDMA, you can have a tradeoff between cell size, users, and data rate. Since the data rate is fixed (sort-of), that leaves a tradeoff between capacity and cell size (noise floor)
Sorry, I don't have my CDMA theory book in front of me, and I don't know the equations off the top of my head. If I get the time, I will look them up for you.
Can anyone else explain to this guy why you can do this in CDMA? Its been too long since I've studied the stuff to get the detail he wants.
I disagree with point 2. IS-95 interoperates better with AMPS. You can have an IS-95 network running on the A frequencies, and standard AMPS on the B frequencies. I'm not sure of this, but it may even be possible to run AMPS and IS-95 in A or B frequencies. The 1.25MHz spectrum of IS-95 is evenly divisable by the channel frequencies of individual AMPS channels.
Many carriers in the U.S. will evolve to CDMA2000, which can be regarded as a stepping stone to WCDMA as soon as they figure out the spectrum. Kind of like how GPRS is a stepping stone from GSM to UMTS.
How long as that nifty Ericsson phone been around? I'm sure its a great phone, but realize that carriers have to plan their networks years before all of these nifty phones come to market.
I'm not knocking GSM...its a great standard and Europe is very fortunate to have a nice uniform standard. It just wasn't right for the U.S. at the time (early 90's). I also like the SIM cards which are non-existant in IS-95.
Does you phone work in central Wyoming, Wisconsin, Montana, Idaho, Arkansas, Oklahoma, New Mexico, or any of the other sparcely populated areas? I guess if you never go to these places, then who cares, huh?
Enough of these ignorant comments.
GSM is, and still isn't right for the United States for 2 reasons.
1) The European GSM standard uses 900MHz and 1800MHz. Those frequencies are used by the United States government, and have been long before GSM came around. So that's why GSM in the U.S. uses different frequencies. So any GSM phone that is, say 2 years old, won't work in Europe and vice versa.
2) Optimal cell size is a function of population density. Digital technologies, especially GSM, require smaller cell sizes. Simply put, places like Wyoming are not going to be getting digital anytime soon. But they do have analog because you can make those cells huge. GSM does not interoperate well with AMPS. CDMA IS-95 does. You can run a CDMA network and an AMPS network at 800MHz.
I also believe that CDMA is the future (not IS-95). Sure, UMTS is based on time-division, but the Docomo 3G call stack is CDMA-based.
So go buy your VoiceStream phone if you live in a big city. I live in Chicago. VoiceStream is there, but no way will I get a VoiceStream phone because if I ever go to Wisconsin, my phone will not get service. My Verizon CDMA phone gets service anywhere in the country.
What is a "cel" site?
At least where I received my B.S. in Electrical Engineering. My senior year, it was like high school all over. All of my classes where in the same building!! In fact, sometimes I didn't even change rooms. No, my school wasn't small being its Big-Ten with over 40,000 students.
So then I went to work for 2 years for a Fortune 50 company and realized that a B.S. in Engineering is just grunt work for several years until you become a manager. So now I'm getting my M.S. or PhD in Computer Science.
Do I wish I had a more well-rounded educations? Hell yes. And I plan to make the most out of Graduate School by taking some off-the-wall classes like Art, Economics, Political Science, and Sociology.
> in the R&D business, that's where all the
> value is
It has been shown that the sweet spot for R&D spending is between 5-15%. Companies that spend more money on R&D inevitable fail. I pulled this statement straight out of Patterson & Hennessey's unpublished third addition of "A Quantitative Approach to Computer Architecture"
Intel is reknown for their manufacturing prowess. Of course the P4 has a low yield. They are just ramping up the Net-burst architecture. Athlon has been running for awhile now.
Personally, I would buy AMD if they could make a chip that doesn't burn up if the CPU fan goes. An Athlon 1GHz uses more than twice the wattage as a 1GHz PIII. Amortise the electricity savings of using a PIII 24/7 for 5 years into the price difference...it may surprise you.
AMD really doesn't stand a chance unless Itanium (IA-64) is a disaster. Intel is betting everything on 64-bit. AMD will never be able to make a chip that uses the Itanium ISA (instruction set architecture). AMD is working on 64-bit extensions to IA-32. However, it will be pointless if IA-64 becomes the IA-32 of the 80's and 90's.
I'll post this again
The ability to overclock a chip says nothing about the quality. It might mean that AMD doesn't have their statistical quality under as tight as control as Intel. They may design a chip to run at 1.6GHz, but because of process inconsistencies, they can only sell them as 1GHz chips. So I ask, are all Athlon's the same? Can every Athlon be clocked the same way or is it a random distribution?
That's basically how it works. A PIII-650 and a PIII-1ghz are the same design, same process, and same chip. However, the more processors they manufacture, the better they get at finetuning their manufacturing/process.
I suggest reading Hennessy & Patterson's "Quantitative Approach to Computer Architecture".
The ability to overclock a chip says nothing about the quality. It might mean that AMD doesn't have their statistical quality under as tight as control as Intel. They may make 1.6GHz chips, but because of process inconsistencies, they can only sell them as 1GHz chips. So I ask, are all Athlon's the same? Can every Athlon be clocked the same way or is it a random distribution?
Sounds like many ideas from the Amoeba and Sprite research operating systems. These were developed in the 80's.
The energy required per transistor is proportional to the product of the load capacitance of the transistor, the frequency of switching, and the square of the voltage.
Can you clarify how decreasing the frequency allows you to lower the voltage? I don't have my VLSI book in front of me right now, but I thought that voltage was proportional to the gate distance...thus smaller processes can operate at a lower voltage.
Well, I don't know what your definition of a mile is. A 1GHz Athlon is faster than a 1GHz PIII, but not by an order of magnitude. I don't have numbers in front of me. Is the SPECint maybe 15% better?
;)
Slowing the clock would not dramatically affect power consumption. Sure, it would help. The Athlon literally uses twice as much power thus dissipating twice as much heat.
So, little boy, when your Athlon burns up because some dust clogged your cooling fan, I'll be happily downloading p0rn with my Intel
You're right! G3/G4's have a superior architecture. I just wish they were commodity hardware like x86...
;)
The x86 has its roots from the 8008 designed in 1974. In fact, the designers of the 8086 spent only a few weeks kludging together an instruction set. Software compatibility means that x86 is still is shitty architecture.
And since the average Joe Blow thinks that MHz is everything, Intel is pushing the P4 to 2GHz even though the SPECint/MHz is far less than the PIII and G4.
Blow on this